yade-dev team mailing list archive
-
yade-dev team
-
Mailing list archive
-
Message #12213
[Bug 1475844] Re: Swap problem in InteractionLoop
Hi Bruno,
I think the two-step solution is right as far as I can see.
For the performance, I apologize for my incorrect statement; you're
right about this. The logic is a bit different in Woo
(https://github.com/eudoxos/woodem/blob/master/pkg/dem/ContactLoop.cpp)
and I will check again what is happening for contacts of undefined shape
combination there (those should be minimized via particle masks, but who
knows).
What I meant is that IIRC the dispatcher lookup is so fast that IIRC it
brings very little to cache the result of the lookup as opposed to
calling the dispatcher directly (in normal circumstances, not when it
repeatedly fails, as in your examples).
Cheers, v.
--
You received this bug notification because you are a member of Yade
developers, which is subscribed to Yade.
https://bugs.launchpad.net/bugs/1475844
Title:
Swap problem in InteractionLoop
Status in Yade:
New
Bug description:
When you use two different materials, it might happen that
IntreactionLoop ends with error [1]. It is because of IPhys stage (see
also the comment in [1], but in reality Ip2 can by non-symmetric).
Possible solutions:
1) define both Ip2_Mat1_Mat2 and Ip2_Mat2_Mat1 (according to [2] maybe does not work..)
2) replace assert [1] (something like if (swap) { I->functorCase.phys=physDispatcher->getFunctor2D(b2->material,b1->material,swap); swap=false }). Is the assert necessary at all?
3) make option 1) somehow automatic (using some macro?)
Pros and cons:
1) would mess the source code and documentation a bit, but otherwise I think it is ok
2) would it have some side effects? Is it ok for Law2 stage? etc etc?
3) maybe the best option if the solution is reasonable
cheers
Jan
[1] http://bazaar.launchpad.net/~yade-pkg/yade/git-trunk/view/head:/pkg/common/InteractionLoop.cpp#L116
[2] https://answers.launchpad.net/yade/+question/269315
To manage notifications about this bug go to:
https://bugs.launchpad.net/yade/+bug/1475844/+subscriptions
Follow ups
References